A New Kind of Index Fund Designed to Grow the Social ...



A New Kind of Index Fund Designed to Grow the Social Security Trust Fund

Dr. David L Wetzell[1]

For presentation at the USBIG Session at EEA 2007

The United States Social Security System is still in need of reform[2]. If it is not provided for, it will be impossible to develop additional income redistribution programs like the Basic Income Guarantee program[3]. But there are serious problems with individual investment accounts as a means to help reduce the deficits in Social Security. Individual investment accounts are not reliable and most individuals are not good at picking stocks[4]. There also are legitimate concerns that smaller investors will be taken advantage of during the ups and downs of the stock market. This paper will not go into these problems, but instead proposes an additional alternative to the privatization of the social security trust fund, the reduction of its benefits or tax policy changes as a means to stabilize the Social Security Trust Fund (SSTF). The alternative considered is the establishment of a public mutual fund, coupled with an effective firewall around the SSTF. The paper outlines an idea for a public mutual fund, as a new kind of index fund, its rationales and illustrates how it would work. The goal of the paper is to motivate the idea as an important progressive reform and the need for its further study, using more historical data and computer model simulations.

The Idea for a Public Mutual Fund

A public mutual fund, referred to as US Mutual, would restore Social Security to sound financial footing, while continuing it as a defined benefit plan that provides forced savings, social insurance and equitable income redistribution. The idea of publicly investing a portion of the SSTF is not a new idea. It was discussed before during the Clinton Administration[5]. The potential contribution is in this paper’s description of a mechanized holdings strategy for US Mutual. The holdings strategy would consider both the common stocks traded on the New York Stock Exchange (NYSE) and the Treasury bills and bonds offered by the US government for its debt. The holdings strategy would form an index of value that would be used to rank stocks and bonds. This is the index that makes the fund a new kind of index fund, with similar low turnover costs[6]. But, unlike with many index funds that limit themselves to holdings within five hundred or four thousand firms, it would allow the consideration of a large number of potential stocks and bonds for investment holdings. The index would be formed in a way that would weight stability more than return. It would then use two-thirds of its available funds to buy long on top-ranked governmental bonds and/or US stocks and one-third of its available funds to sell short on bottom-ranked US stocks[7]. It would be from selling short on over-valued unstable under-performing stocks that US Mutual would grow the Social Security Trust Fund.

The Intended Purposes of a Mechanized Holding Strategy

However, there are additional reasons for the mechanized holding strategy that will be described below. These reasons distinguish why the investment strategy would differ from a corporate or private model for investment in stocks. Its purposes are as follows:

1) To improve on the return on the Social Security Trust Fund in a manner that is meritocratic, has low overhead costs, and hedges well against a worse-case scenario of a general downturn in the US economy.

2) To constrain the effects of US Mutual’s market power.

3) To promote low turnover in its long-holdings of stocks.

4) To diversify US Mutual in a transparent manner.

5) To reduce the volatility in the New York Stock Exchange.

6) To promote the stability of the US Economy and US Treasury Bills and to motivate the reduction of US debt.

These purposes will be clarified as the holdings strategy is explained below.

An Overview of the Mechanized Holdings Strategy

The basic strategy would be to use twenty-seven weeks of publicly available weekly data and an easy to replicate algorithm to estimate predicted weekly log-returns and their standard deviations for all the stocks and bonds considered[8]. The use of public data and a relatively simple algorithm is what ensures transparency. This is because those people with a vested interest in holding US Mutual accountable would be able to use the same data and algorithm to verify the integrity of the fund. The predicted values of the log-return and its standard deviation for all stocks and bonds considered by US Mutual would then be combined into an index of value. The index of value would be based on a simple formula. It would be found by taking one-third times the predicted weekly log-return minus two-thirds times the predicted standard deviations. This index embodies a conservative investment strategy that prioritizes stability over return. Such an investment strategy is mandated when one considers that what is being invested constitutes a substantial fraction of many people’s retirement income. The index would then be used to determine the weekly holding targets of US Mutual.

We must digress some before getting into the details on how the weekly holding targets would be chosen. As mentioned before, the mechanized holding strategy dedicates two-thirds of the available portion of the SSTF to buying long on T-bills/bonds and stocks. The top forty percent of the weighted combined value of bonds and/or stocks, weighted by their predicted total values, would be held long by US Mutual. The bottom ten percent of the weighted combined value of the stocks and bonds would be sold short by US Mutual[9]. The fraction of forty percent is chosen based on the prediction that the index would naturally tend to value government bonds more than stocks[10]. According to Wikipedia, “the New York Stock Exchange has a global capitalization of $17.4 trillion, including $7.1 trillion in non-U.S. companies.” If US Mutual were required to hold only US companies then the total value of the NYSE would be approximately ten trillion[11]. The size of the US debt, as of 2006, when one includes intragovernment debt obligations was 8.6 trillion dollars. This implies that total government debt would currently be more than forty percent of the sum of the total government debt and the total market valuation of the NYSE. It also implies that US Mutual’s long-holdings would initially be completely in government bonds and t-bills. This could change when the size of the government debt is relatively reduced. However, given the historical persistence of large government debts in the US, the general tendency would be to invest a relatively small fraction of the SSTF in long-holdings in the NYSE. As mentioned before, this implies a reliance on US Mutual’s selling short strategy and, of course, a much better lock box for the SSTF, for the growth of the SSTF.

We can now return to how the index values are used to determine the targeted holdings of US Mutual. Let us introduce some mathematical notation to help with the description of the algorithm. For the sake of simplicity, all of the possible stocks and bonds considered by US Mutual are assigned a number i. We will discriminate different i from each other by whether US Mutual would buy long or sell short on them. Let us say that l is the subset that US Mutual would buy long on and s is the subset that US Mutual would sell short on. So when we want to refer to stocks sold short, we would consider every [pic]and when we want to refer to bonds or stocks bought long, we would consider every [pic]. Now, the information that US Mutual would use to determine its holdings include the total predicted total valuations of individual stocks or bonds, or Mi, the combined total value of the NYSE and the US Government debt, T, the size of the available portion of the SSTF, F, and the index values, vi, of the stocks or bonds considered by US Mutual[12]. Now, to determine the short-selling holdings, the [pic] are linearly transformed to [pic] where the weighted average, using the Mi as weights, of all of the[pic]in s is one. The targeted amount sold short for stock i would then be [pic]. The targeted amount bought long for bond or stock i would then be [pic].

So now we can show the average percentage holdings in bonds and the percentage of a stock sold short. The size of the SSTF at the end of 2006 was about two trillion[13]. Let us presume that the available portion, or F, is one trillion. Then, if M is eighteen trillion, the average percentage of stock sold short would be 18.5 % and US Mutual would hold onto 9.3 % percent of the US government’s debt instruments.

Some Additional Assumptions made.

This strategy is complemented well by changing Social Security system from the pay-as-you go method to a process known as Actuarial Advance Funding (AAF) and, more specifically an actuarial cost method known as Entry Age Normal[14]. This method involves carefully forecasting future payments and regularly and methodically setting aside funds so that the remainder of funds received could be safely invested. The details of the actuarial method are not critical for this paper. What matters is the implication that a variable amount of the SSTF, or F, would be available for potential investment[15].

This strategy also presumes that the current legal restrictions on selling short would not apply to US Mutual[16]. These regulations are presumed to be worthwhile in large part because of the potential effectiveness of market power in leveraging the value downwards. US Mutual is implicitly here given a monopoly role as the official “bottom-feeder” of the NYSE.

A Defense of the Benefits of US Mutual.

It would serve to bleed over-valued stocks that are unstable under-performers. It may also help to discourage bubbles and trading behavior that tries to time the market and, in doing so, tends to make the stock market overall more volatile. The reduced overall volatility in the stock market would be beneficial for smaller investors and companies. Any US company that would enter its stock on the NYSE would have a full twenty-seven weeks before their stock would come under the scrutiny of US Mutual. There would likely be a transition period from the start of US Mutual, during which some naturally more volatile and less profitable stocks may get forced out of the NYSE. However, eventually, the market would adjust to US Mutual as an uber-institutional investor. When over-valued stocks are sold short enough so that they are under-valued the market will reverse itself. Although, since the stock’s value is lower and its recent history makes the threat of US Mutual selling short on it again more viable the price would not rise to its previous level. This is what would allow US Mutual to grow the SSTF, by selling (short when) high and buying (back when) low.

At the face of it, this might reduce the total value of the NYSE and drive some corporations away from the NYSE, but if corporations react by finding ways to reduce the volatility of their stocks then the overall investment climate in the NYSE would be changed with some significant consequences. A less volatile stock market could attract more investments from smaller investors and corporations that had previously held their assets in land, bonds or other more stable assets. These new investors, as well as previous investors, would hopefully plan their holdings carefully with lower rates of turnover. This would help to both increase the total value of the NYSE and further stabilize it as a market. This would then affect the price of bonds and increase the equity premium for stocks, so that corporations that have a higher equity-to-debt ratio would be favored. These implications would tend to make the NYSE more accessible to the general public and grow the SSTF without additional taxes or reduced benefits.

Another key benefit for the mechanized holdings policy would be in how it would protect against the politicization and corruption of the US Mutual. The potential for corruption is particularly important when one considers how the size of US Mutual would easily dwarf other private mutual funds, like the over a 100 billion dollars that the Vanguard 500 invests[17]. Although, US Mutual would not compete with private mutual funds, since its focus on stability rather than return would be quite different from other mutual funds. But it would wield considerable market power, which is why the potential for corruption is a key concern for a public mutual fund. The chief way to mitigate the importance of market power is to consider the US government as a host of companies with its treasury bills or bonds treated as types of stocks. But, if in the future, the US debt were reduced, then US Mutual would also buy long in stable performers in the NYSE. These stocks would have no chance of influencing the fund-managers, who may face principal-agent problems, since the holdings would be determined completely by the algorithm described above.

A serious concern that bothers many people with the US government investing their social security savings in the stock market is the worse case scenario of a general decline in the stock market, not unlike what happened with the Great Depression or recent stock market declines. This fear is unfounded for US Mutual, since US Mutual would likely never invest a very large portion of its available funds in long-holdings in the NYSE. The long-holdings it did have would be in the most stable stocks, like Warren Buffett’s Berkshire Hathaway. These stocks would likely have lower betas, or correlation with the overall stock markets. As such, with a general decline, they would decline by less and the fact that US Mutual would hold around ten percent or more of their stock’s value would further increase the stability of the stock in the face of an overall decline. The holdings in government bonds and short-holdings of US Mutual, whose betas would tend to be higher than average, would generally benefit from a general decline in the NYSE, so US Mutual would have a very low beta and safeguard the SSTF well against the worse case scenario of a general decline in the NYSE. Thereby, it would help to improve the overall macro-stability of the US economy.

US Mutual’s algorithm for its holdings would tend to be reinforcing, despite the existence of significant market power. In other words, when US Mutual would buy long a significant share of a stock, its log-return may decline somewhat as a result, but its standard deviation would also decline and the net effect would be to neutralize or increase the index value of the stock. In equilibrium, the short-selling strategy would tend to have a higher turnover rate so that stocks would not be bled dry, but rather brought down to around their proper value. In doing this, US Mutual would be capitalizing on the effects of the long-term planning of others and helping the stock market to work better.

Some Preliminary Evidence about How it would Work.

But the above are speculations. They need to be corroborated through thorough research using historical data and computer simulations. This paper provides preliminary evidence of how it would work using monthly stock market data from the S&Ps of Mexico’s Bolsa stock market[18]. This is a time series dataset, not a panel dataset with several different stocks. It serves to illustrate how the predictions of log-returns and their standard deviations can be combined into a stable index of value over time.

The logs of the stock values are taken and the log-return is calculated by first a first-difference of the log-values of the stock-values[19]. Six consecutive log-return observations are used with a mean regression to predict the log-return for the next month, mnt. The absolute values of the residuals from the log-return regressions are then used, with the same sort of regression, to predict the standard deviation of the log-return, mnsdt. The index of value is then formed as [pic]. You can see the descriptive statistics of the stock values, their log-values, their log-returns, the predicted means and standard deviations and the index values in table one. It is apparent that the predicted values and the index values are far more stable than the observed log-returns. What is interesting is that the index values are more stable than both the predicted means and standard deviations. The index values are somewhat more stable overall than the standard deviations, which are more stable than the means. The skew and kurtosis statistics are of interest. It is important to remember that the skew of symmetric distribution is zero and the kurtosis of a normal distribution is three. A higher kurtosis means that the distribution’s tails are unusually high. The predicted standard deviations are subject to being somewhat more skewed and with fatter tails than the predicted means. The index value is similar to the predicted standard deviations, but somewhat less extreme.

These are just descriptive statistics. What really matters is to see how stable the values are over time. To estimate this, a median autoregressive of order one regression was used to analyze the predicted mean and standard deviation and the index value. A median regression was used because of the non-normal skew and kurtosis in the standard deviations and index value. The correlation between the index values was very high at .97. This is significantly better than the other two variables, with a greater accuracy as well as shown by the lower standard errors and the regression’s higher R-squared. The variable does not have unit root, but it comes relatively close, meaning its value in month t tends to be very close to its value in month t-1. You can also get a sense of how the strategy would work in graph one, where the observed log-returns in the last nine years of the dataset are contrasted with the index values.

This is absolutely critical for ensuring the stability of US Mutual’s holdings, as the index values are what would be used to determine holdings. Given how the predicted standard deviations are more stable than the predicted log-returns, the conservative holding strategy that weights standard deviations twice as much as predicted log-returns is key for ensuring that the mechanized holding strategy would have low-turnover. This would help keep US T-bills, and eventually US stock values more stable.

Conclusion

The paper has sketched out the rationales for a public mutual fund as the key means to stabilize the Social Security Trust Fund. It has described the purposes of such a fund, detailed a mechanized holding strategy that would make it behave similar to an index fund and protect against its corruption. It has shown how the public mutual fund would, with the help of a fire-wall around the Social Security Trust Fund, combine holdings of mainly US Treasury bills or bonds with an innovative strategy to sell short on unstable under-performing stocks. It gives a taste of how and why the algorithm would work. But it is only a taste and more extensive work is needed using historical data and computer simulations. Of these two, the computer simulations would be the more important part because they would be needed to test out the selling short strategy. One would particularly need to make realistic behavioral assumptions about the impact of the use of US Mutual’s market power in selling short stocks, in both the short-run and the long-run[20].

This is an idea, different from the main ideas in the policy arena about how to secure the future of Social Security and other suggestions about integrating an investment in the stock market. If anything, it verifies Barr(2002)’s conclusion that the key variable is effective government. It takes an effective government to develop the will to set the Social Security Trust Fund aside with an effective firewall and to allow it a monopoly in systematically bleeding under-performing stocks in the NYSE as the way to provide for the retirement of the Baby Boomer generation. This is a critical step to allow other reforms, like a Basic Income Guarantee that would also effectively improve on how the US redistributes income in a non-discretionary manner that imposes more discipline on our public governance.

Bibliography

Barr, Nicholas. “Reforming Pensions: Myths, truths, and policy choices” International Social Security Review 55(2) 3-36.

Century Foundation, The. “Investing the Social Security Trust Funds in Stocks” 3/1/1999 as found at < > on 7/21/06.

Diamond, Peter A., Orszag, Peter R. Saving Social Security: A Balanced Approach

Google Books, Brookings Institution, 2005.

Maynard, John Kenneth. A Tenured Professor New York, Mariner Books, 2001.

Mitchell, Olivia S. & Zeldes, Stephen P. “Social Security Privatization: A structure for Analysis” Vol 88, No 2, Papers and Proceedings of the Hundredth and Eight Annual Meeting of the AEA San Francisco, CA, January 5-7, 1996 (May, 1996), p 363-367.

Modigliani, Franco and Miller, Merton H. “The Cost of Capital, Corporation Finance and the Theory of Investment” The American Economic Review, Vol. 48, No. 3 (Jun., 1958), pp. 261-297

Table 1

Descriptive Statistics of Mexican Bolsa S&P500

| |Stock Values |Ln (Stock |100*Log-Returns |Predicted |Predicted |Index Values |

| | |Values) | |Means of |Std Dev | |

| | | | |100Log-Returns |100Log-Returns | |

|Sample Size |558 |558 |557 |551 |551 |551 |

|Mean |118.3 |4.42 |.607 |.613 |2.28 |-1.31 |

|(Std Dev) |(103.4) |(.87) |(3.37) |(1.55) |(1.03) |(.99) |

|Skew |1.57 |-.17 |-.63 |-.38 |1.20 |-1.07 |

|Kurtosis |4.66 |2.58 |4.72 |3.22 |5.48 |5.21 |

Table 2

A Median Autoregressive Time Series Analysis of Order One of

the Predicted Mean, Standard Deviation and Index Values.

| |Predicted Mean* |Predicted Std Dev* |Index Value* |

|Lagged Value |.912 |.936 |.972 |

| |(.022) |(.015) |(.007) |

|Constant |.059 |.137 |-.040 |

| |(.037) |(.037) |(.012) |

|R-squared |.57 |.56 |.62 |

Graph 1

A comparison of the observed Log-Returns with Index Values.

[pic]

Graph 2: Markowitz Frontier[21]

[pic]

Appendix 1

Where the Idea Came From?

The idea for US Mutual’s mechanized holding strategy is loosely based on Modigliani and Miller (1958) model for the portfolio selection based on the expected return and standard deviation[22]. In graph two, a diagram of risk-return space is given. That is it has the standard deviation on the x-axis and the expected return on the y-axis. A smoothly bordered convex set is shown that contains within its borders all of the possible combinations of returns and standard deviations for possible investment portfolios. Modigliani et al (1958) then presume the theoretical existence of a risk-free asset and an optimal portfolio on the markowitz frontier, or the upper edge of the set of possible portfolios. A straight line is drawn between the risk-free asset and the optimal portfolio. An investor could then choose what combination of risk-free assets and stock investments they desire based on their willingness to accept more risk for a better return on their savings. This model is based on the presumption that the investor is relatively small so their portfolio selection strategies do not impact the return or standard deviation. It also presumes that there are no problems with trust or principle-agent problems, i.e. where the fund managers who determine the holdings in an investment portfolio are not the ultimate benefactors of the investments. It also does not delve into the thorny practical questions of how one calculates or projects the expected stock returns or their expected standard deviations[23]. The virtue of the model is its simplicity, but the above aspects of “reality” that it abstracts from provide for the earlier mentioned rationales for the mechanized holding strategy.

-----------------------

[1] Currently affiliated with Bethel University and Seminary, St Paul, MN 55112.

[2] This is almost the same sentence as the introductory sentence of Olivia S Mitchell and Stephen P Zeldes’s paper “Social Security Privatization: A structure for Analysis” Vol 86, No. 2, Papers and Proceedings of the Hundredth and Eight Annual Meeting of the AEA San Francisco, CA, January 5-7, 1996 (May, 1996), pp 363-367.

[3] General information about the Basic Income Guarantee program is available at the website.

[4] This paper has been helped by internet interaction with the wisdom of actuary Andy Lang. He goes into why it is not wise to privatize social security at his blog, an excerpt of which was reposted at < > on July 25, 2006.

[5] The Century Foundation, “Investing the Social Security Trust Funds in Stocks” 3/1/1999 as found at < > on 7/21/06.

[6] Index Funds are passively managed, limited-expense (management fee no higher than 0.50%) mutual funds designed to replicate the performance of an unmanaged stock market index. It is worth noting that the turnover costs would be reduced for US Mutual as they would not need to pay capital gains taxes.

[7] US Mutual, as a public mutual fund, would be exempt from current legal restrictions on selling short in the NYSE.

[8] Log-return is the ln(1+return). A weekly return is the ratio between the current end-of-week value and last week’s end-of-week value. The simplest way to generate the predicted log-returns and their standard deviations would be from a mean regression of the previous twenty-six log-returns. The residuals from these regressions would then be used to predict the standard deviations. The easiest way to do this would be to take the absolute values of the twenty-six residuals and use them in a simple mean regression to predict the standard deviation.

One could also conceivably allow for an autoregressive of order one regression to be used in the prediction of log-returns and their standard deviations, but this is an alternative that is not discussed in this paper. Another approach that could be used to predict the log-returns and their standard deviations would be to use median instead of mean regressions. These alternative approaches could hopefully be tested out in further empirically oriented studies on the matter.

[9] The ten percent is chosen so as to reduce the number of firms that would be sold short on and to increase the intensity of the market power to on average twice the level of buying long, as market power will be critical for the selling short strategy elaborated later.

[10] The index weights the standard error twice as much as the log-return. As such, since government bonds tend to approximate risk-free investments as closely as possible with quite low standard errors, the mechanized algorithm would favor them.

[11] The rationales for limiting the stock holdings of US Mutual would be based on the positive externalities of economic stability for the US’s economy and the creation and sustainment of US jobs, as well as reduced turnover in holdings from fewer stocks overall being considered.

[12] The predicted total market capitalization of stocks would be predicted from the same public data, using their past history and a linear trend. The same technique would be used for US Treasury Bonds. This implicitly treats the US government as like a stock, but one whose return is relatively risk free.

[13] Social Security Trust Funds, as found at < > on February 9, 2007.

[14]Information on Entry Age Normal can be found at google’s online books < >.

[15] This also presumes that a truly secure lock-box is set around the SSTF. This is a nontrivial matter on how to politically commit not to raid the SSTF, but we are presuming that an extended commitment to respect the SSTF would be feasible.

[16] For a description of the current regulations and their effects, see < > as found on February 9, 2007.

[17] Vanguard Group, Wikipedia as found at < > on February 9, 2007.

[18] The data includes the market valuations of the S&P 500 from January of 1947 to June of 1993. This is a total of 558 observations of data.

[19] A first difference is formed when a time series variable[pic]is lagged one time period to form [pic]and [pic]is calculated.

[20] Where the short run is where the stocks need to readjust to the existence of US Mutual and some quite volatile stocks may leave the NYSE because of US Mutual. And where the long run is where the stocks have adjusted to the existence of US Mutual by adopting strategies that will reduce their volatility and rescue overvalued stocks when they are under their proper value.

[21] Image: Markowitz Frontier. As found at < > on 2/10/07.

[22] It also derives some inspiration from the satirical novel “A Tenured Professor” by James K. Galbraith, where an economics professor devises a strategy to make money off the stock market by systematically selling short on overvalued stocks.

[23] These questions touch on deeper philosophical issues that surround the meaning of probability and statistical measurement. For an overview of these issues, consider the entries on probability in the New Palgrave’s dictionary of Economics.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download